[Bug 40963] New: Upward, lonely robot Demo crashes with Too many buffers queued

https://bugs.winehq.org/show_bug.cgi?id=40963

            Bug ID: 40963
           Summary: Upward, lonely robot Demo crashes with Too many
                    buffers queued
           Product: Wine
           Version: 1.9.14
          Hardware: x86
                OS: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: x3daudio
          Assignee: wine-bugs(a)winehq.org
          Reporter: leslie_alistair(a)hotmail.com
      Distribution: ---

Created attachment 55110
  --> https://bugs.winehq.org/attachment.cgi?id=55110
Log of +xaudio2

Running Upward, lonely robot Demo, crash after it detects "Too many buffers
queued".

Log attached.

-- 
Do not reply to this email, post in Bugzilla using the
above URL to reply.
You are receiving this mail because:
You are watching all bug changes.

[Bug 40963] Upwards, lonely robot Demo crashes with Too many buffers queued

https://bugs.winehq.org/show_bug.cgi?id=40963

Alistair Leslie-Hughes <leslie_alistair(a)hotmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|Upward, lonely robot Demo   |Upwards, lonely robot Demo
                   |crashes with Too many       |crashes with Too many
                   |buffers queued              |buffers queued

-- 
Do not reply to this email, post in Bugzilla using the
above URL to reply.
You are receiving this mail because:
You are watching all bug changes.

[Bug 40963] Upwards, lonely robot Demo crashes with Too many buffers queued

https://bugs.winehq.org/show_bug.cgi?id=40963

--- Comment #1 from Alistair Leslie-Hughes <leslie_alistair(a)hotmail.com> ---
Created attachment 55111
  --> https://bugs.winehq.org/attachment.cgi?id=55111
A possible fix

The issue is caused by nbufs being decreased without check if it's already 0.

With this applied the game starts (assuming it doesn't hit a texture bug), and
is quite playable.

-- 
Do not reply to this email, post in Bugzilla using the
above URL to reply.
You are receiving this mail because:
You are watching all bug changes.

[Bug 40963] Upwards, lonely robot Demo crashes with Too many buffers queued

https://bugs.winehq.org/show_bug.cgi?id=40963

Alistair Leslie-Hughes <leslie_alistair(a)hotmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |patch
                 CC|                            |aeikum(a)codeweavers.com

--- Comment #2 from Alistair Leslie-Hughes <leslie_alistair(a)hotmail.com> ---
Adding Andrew for comment before upstreaming.

-- 
Do not reply to this email, post in Bugzilla using the
above URL to reply.
You are receiving this mail because:
You are watching all bug changes.

[Bug 40963] Upwards, lonely robot Demo crashes with Too many buffers queued

https://bugs.winehq.org/show_bug.cgi?id=40963

--- Comment #3 from Andrew Eikum <aeikum(a)codeweavers.com> ---
(In reply to Alistair Leslie-Hughes from comment #1)
> Created attachment 55111 [details]
> A possible fix
>
> The issue is caused by nbufs being decreased without check if it's already 0.
>
> With this applied the game starts (assuming it doesn't hit a texture bug),
> and is quite playable.

This seems like a workaround instead of a proper fix.

I think a better solution would be to clear the OpenAL source object's
processed buffers during Stop, calling callbacks as needed, then drop any
remaining buffers in the source. This should prevent nbufs from being
decremented after the flush call, since there should be no more buffers
remaining from before the flush.

-- 
Do not reply to this email, post in Bugzilla using the
above URL to reply.
You are receiving this mail because:
You are watching all bug changes.

[Bug 40963] Upwards, lonely robot Demo crashes with Too many buffers queued

https://bugs.winehq.org/show_bug.cgi?id=40963

--- Comment #4 from Alistair Leslie-Hughes <leslie_alistair(a)hotmail.com> ---
(In reply to Andrew Eikum from comment #3)
> I think a better solution would be to clear the OpenAL source object's
> processed buffers during Stop, calling callbacks as needed, then drop any
> remaining buffers in the source. This should prevent nbufs from being
> decremented after the flush call, since there should be no more buffers
> remaining from before the flush.

Did you want to write it?  Happy to test..

-- 
Do not reply to this email, post in Bugzilla using the
above URL to reply.
You are receiving this mail because:
You are watching all bug changes.

[Bug 40963] Upwards, lonely robot Demo crashes with Too many buffers queued

https://bugs.winehq.org/show_bug.cgi?id=40963

--- Comment #5 from Andrew Eikum <aeikum(a)codeweavers.com> ---
(In reply to Alistair Leslie-Hughes from comment #4)
> (In reply to Andrew Eikum from comment #3)
> > I think a better solution would be to clear the OpenAL source object's
> > processed buffers during Stop, calling callbacks as needed, then drop any
> > remaining buffers in the source. This should prevent nbufs from being
> > decremented after the flush call, since there should be no more buffers
> > remaining from before the flush.
>
> Did you want to write it?  Happy to test..

I will when I find time, but I won't be sad if you beat me to it :)

-- 
Do not reply to this email, post in Bugzilla using the
above URL to reply.
You are receiving this mail because:
You are watching all bug changes.

[Bug 40963] Upwards, lonely robot Demo crashes with Too many buffers queued

https://bugs.winehq.org/show_bug.cgi?id=40963

--- Comment #6 from Andrew Eikum <aeikum(a)codeweavers.com> ---
Created attachment 55163
  --> https://bugs.winehq.org/attachment.cgi?id=55163
xaudio2: Ignore queued AL buffers after Stop

I'm not thrilled with this patch, but I think it will fix it.

I wasn't able to find a way to remove buffers from the alSource before they're
played. So instead we just ignore them as they're completed playing by OpenAL.

-- 
Do not reply to this email, post in Bugzilla using the
above URL to reply.
You are receiving this mail because:
You are watching all bug changes.

[Bug 40963] Upwards, lonely robot Demo crashes with Too many buffers queued

https://bugs.winehq.org/show_bug.cgi?id=40963

--- Comment #7 from Alistair Leslie-Hughes <leslie_alistair(a)hotmail.com> ---
(In reply to Andrew Eikum from comment #6)
> Created attachment 55163 [details]
> xaudio2: Ignore queued AL buffers after Stop
>
> I'm not thrilled with this patch, but I think it will fix it.
>
> I wasn't able to find a way to remove buffers from the alSource before
> they're played. So instead we just ignore them as they're completed playing
> by OpenAL.

Thanks,  confirming that it also fixes the crash.

-- 
Do not reply to this email, post in Bugzilla using the
above URL to reply.
You are receiving this mail because:
You are watching all bug changes.

[Bug 40963] Upwards, lonely robot Demo crashes with Too many buffers queued

https://bugs.winehq.org/show_bug.cgi?id=40963

Alistair Leslie-Hughes <leslie_alistair(a)hotmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
  Attachment #55111|0                           |1
        is obsolete|                            |

-- 
Do not reply to this email, post in Bugzilla using the
above URL to reply.
You are receiving this mail because:
You are watching all bug changes.

[Bug 40963] Upwards, lonely robot Demo crashes with Too many buffers queued

https://bugs.winehq.org/show_bug.cgi?id=40963

Sebastian Lackner <sebastian(a)fds-team.de> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
    Staged patchset|                            |https://github.com/wine-com
                   |                            |pholio/wine-staging/tree/ma
                   |                            |ster/patches/xaudio2_7-Aban
                   |                            |don_Albufs
             Status|NEW                         |STAGED
                 CC|                            |dmitry(a)baikal.ru,
                   |                            |erich.e.hoover(a)wine-staging
                   |                            |.com, michael(a)fds-team.de,
                   |                            |sebastian(a)fds-team.de

-- 
Do not reply to this email, post in Bugzilla using the
above URL to reply.
You are receiving this mail because:
You are watching all bug changes.

[Bug 40963] Upwards, lonely robot Demo crashes with Too many buffers queued

https://bugs.winehq.org/show_bug.cgi?id=40963

--- Comment #8 from Sebastian Lackner <sebastian(a)fds-team.de> ---
(In reply to Andrew Eikum from comment #6)
> Created attachment 55163 [details]
> xaudio2: Ignore queued AL buffers after Stop
>
> I'm not thrilled with this patch, but I think it will fix it.
>
> I wasn't able to find a way to remove buffers from the alSource before
> they're played. So instead we just ignore them as they're completed playing
> by OpenAL.

Based on Alistairs request I've added it to Wine Staging anyway. Please note
that the patch attached here leads to compiler warnings (unused variables) and
contains mistake in the handling of abandoned_albufs.
> if(processed > src->abandoned_albufs){
>            processed -= src->abandoned_albufs;
>            i = src->abandoned_albufs;

Setting i here has no effect.
> src->abandoned_albufs = 0;
>
> src->first_al_buf += processed;
>            src->first_al_buf %= XAUDIO2_MAX_QUEUED_BUFFERS;
>            src->al_bufs_used -= processed;
>
> for(i = 0; i < processed; ++i){

This will iterate over the wrong range of buffers.

Those issues are fixed in the patch included in Wine Staging.

-- 
Do not reply to this email, post in Bugzilla using the
above URL to reply.
You are receiving this mail because:
You are watching all bug changes.

[Bug 40963] Upwards, lonely robot Demo crashes with Too many buffers queued

https://bugs.winehq.org/show_bug.cgi?id=40963

Anastasius Focht <focht(a)gmx.net> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |focht(a)gmx.net

--- Comment #9 from Anastasius Focht <focht(a)gmx.net> ---
Hello folks,

since this is the only ticket that references this staged patchset:

https://github.com/wine-compholio/wine-staging/blob/master/patches/xaudio2_7...

This patch breaks World of Warships 0.5.16.x and earlier from Wargaming - it
crashes shortly after start.
Since Wine-Staging is the only way to get decent graphics performance out of
the game (CSMT), I've had the patch on my ignore list for months and forgot to
report it.
Actually I was wondering why no one else reported it the whole time.

--- snip ---
$ pwd
/home/focht/wine-games/wineprefix64-wargaming/drive_c/Games/World_of_Warships

$ wine ./WorldOfWarships.exe
...
Unhandled exception: page fault on read access to 0x00000054 in 32-bit code
(0x0138e860).
Register dump:
 CS:0023 SS:002b DS:002b ES:002b FS:0063 GS:006b
 EIP:0138e860 ESP:1648e8dc EBP:1648e918 EFLAGS:00210297(  R- --  I S -A-P-C)
 EAX:00000004 EBX:f74d2000 ECX:00000054 EDX:12ba4d34
 ESI:14e80cb4 EDI:1648fb40

Backtrace:
=>0 0x0138e860 in worldofwarships (+0xf8e860) (0x1648e918)
  1 0xf3879396 do_engine_tick+0x232(This=0x211b80)
[/home/focht/projects/wine/wine.repo/staging-build-x86/dlls/xaudio2_7/../../include/xaudio2.h:5965]
in xaudio2_7 (0x1648e9c8)
  2 0xf387966b engine_threadproc+0x94(arg=<couldn't compute location>)
[/home/focht/projects/wine/wine.repo/staging-src/dlls/xaudio2_7/xaudio_dll.c:2454]
in xaudio2_7 (0x1648e9f8)
  3 0x7bc972d4 call_thread_func_wrapper+0xb() in ntdll (0x1648ea28)
  4 0x7bc9730f call_thread_func+0x30(entry=0xf38795d6, arg=0x211b80,
frame=0x1648eb28)
[/home/focht/projects/wine/wine.repo/staging-src/dlls/ntdll/signal_i386.c:2815]
in ntdll (0x1648eb08) 
...
0x0138e860: movl    0x0(%ecx),%eax
Modules:
Module    Address            Debug info    Name (204 modules)
PE      340000-  38c000    Deferred        ortp
PE      390000-  3c4000    Deferred        ilu_x86r
PE      3d0000-  3de000    Deferred        gettext_x86r
PE      400000- 277c000    Export          worldofwarships
PE     2780000- 2c7c000    Deferred        vivoxsdk
PE     2c80000- 2ed3000    Deferred        resil_x86r
PE    10000000-101ff000    Deferred        d3dx9_43
ELF    7a800000-7a942000    Deferred        opengl32<elf>
  \-PE    7a840000-7a942000    \               opengl32
ELF    7b400000-7b7ec000    Deferred        kernel32<elf>
  \-PE    7b420000-7b7ec000    \               kernel32
ELF    7bc00000-7bd11000    Dwarf           ntdll<elf>
  \-PE    7bc30000-7bd11000    \               ntdll
ELF    7c000000-7c004000    Deferred        <wine-loader>
...
Threads:
process  tid      prio (all id:s are in hex) 
...
00000091 WargamingGameUpdater.exe
    00000095    0
    00000092    0
000000b0 (D) C:\Games\World_of_Warships\worldofwarships.exe
    000000c7    1
    000000c6    0
    000000c5   15
    000000c4    0 <==
    000000c3    0
    000000c2    0
    000000c1    1
    000000c0    0
    000000bf    0
    000000be    0
    000000bd    0
    000000bc    0
    000000bb    0
    000000ba    0
    000000b9    0
    000000b8    0
    000000b7    0
    000000b6    0
    000000b1    0 
...
System information:
    Wine build: wine-2.0-rc3-900-g9b4c5ab (Staging)
    Platform: i386 (WOW64)
    Version: Windows XP
    Host system: Linux
    Host version: 4.4.14-200.fc22.x86_64
--- snip ---

Better trace:

--- snip ---
$ WINEDEBUG=+tid,+seh,+relay,+xaudio2 wine ./WorldOfWarships.exe >>log.txt 2>&1
...
0123:Call KERNEL32.LoadLibraryExW(0032e45e
L"C:\\windows\\system32\\xaudio2_7.dll",00000000,00000008) ret=014a19d6
0123:Call PE DLL (proc=0xf3c25695,module=0xf3c10000
L"xaudio2_7.dll",reason=PROCESS_ATTACH,res=(nil))
0123:trace:xaudio2:DllMain (0xf3c10000, 1, (nil))
...
0123:Ret  PE DLL (proc=0xf3c25695,module=0xf3c10000
L"xaudio2_7.dll",reason=PROCESS_ATTACH,res=(nil)) retval=1
0123:Ret  KERNEL32.LoadLibraryExW() retval=f3c10000 ret=014a19d6 
...
0123:trace:xaudio2:XAudio2CF_CreateInstance
(0xfc87548)->((nil),{8bcf1f58-9fe7-4583-8ac6-e2adc465c8bb},0x32e7dc)
...
0123:trace:xaudio2:IXAudio2Impl_QueryInterface
(0xfc87560)->({8bcf1f58-9fe7-4583-8ac6-e2adc465c8bb}, 0x32e7dc)
0123:trace:xaudio2:IXAudio2Impl_AddRef (0xfc87560)->(): Refcount now 1 
...
0123:trace:xaudio2:IXAudio2Impl_StartEngine (0xfc87560)->()
0123:Call
KERNEL32.CreateThread(00000000,00000000,f3c255d6,0fc87560,00000000,00000000)
ret=f3c239d2
0123:Ret  KERNEL32.CreateThread() retval=000001a8 ret=f3c239d2
0123:trace:xaudio2:XAudio2CF_CreateInstance Created XAudio version 27:
0xfc87560
0123:trace:xaudio2:XAudio2CF_Release (0xfc87548)->(): Refcount now 0
...
0123:Ret  ole32.CoCreateInstance() retval=00000000 ret=013fcc69
0123:trace:xaudio2:XA27_Initialize (0xfc87560)->(0x0, 0xffffffff) 
...
0133:Starting thread proc 0xf3c255d6 (arg=0xfc87560)
...
0123:trace:xaudio2:XA27_CreateMasteringVoice (0xfc87560)->(0x14e30cc4, 0,
48000, 0x0, 0, (nil))
0123:trace:xaudio2:IXAudio2Impl_CreateMasteringVoice (0xfc87560)->(0x14e30cc4,
0, 48000, 0x0, L"{0.0.0.00000000}.{FD47D9CC-4218-4135-9CE2-0C195C87405B}",
(nil), 0x6)
... 
0123:trace:xaudio2:IXAudio2Impl_RegisterForCallbacks (0xfc87560)->(0x14e30cb0)
0123:trace:xaudio2:XA27_GetDeviceDetails 0xfc87560, 0, 0x32e940 
...
0133:trace:xaudio2:do_engine_tick frames available: 1440
...
0123:trace:xaudio2:IXAudio2Impl_CreateSourceVoice (0xfc87560)->(0x14e30cc8,
0x32ed6c, 0x0, 1.000000, 0x14e30cb4, (nil), (nil))
0123:trace:xaudio2:dump_fmt wFormatTag: 0xfffe (WAVE_FORMAT_EXTENSIBLE)
0123:trace:xaudio2:dump_fmt nChannels: 2
0123:trace:xaudio2:dump_fmt nSamplesPerSec: 48000
0123:trace:xaudio2:dump_fmt nAvgBytesPerSec: 384000
0123:trace:xaudio2:dump_fmt nBlockAlign: 8
0123:trace:xaudio2:dump_fmt wBitsPerSample: 32
0123:trace:xaudio2:dump_fmt cbSize: 22
0123:trace:xaudio2:dump_fmt dwChannelMask: 00000003
0123:trace:xaudio2:dump_fmt Samples: 0020
0123:trace:xaudio2:dump_fmt SubFormat: {00000003-0000-0010-8000-00aa00389b71} 
...
0133:Ret  KERNEL32.WaitForSingleObject() retval=00000000 ret=f3c255ff
...
0123:trace:xaudio2:XA2SRC_SetOutputVoices 0xfc8c9d8, (nil)
0123:trace:xaudio2:XA2SRC_SetOutputVoices Outputting to: 0x0, 0xfc87564
...
0133:trace:xaudio2:do_engine_tick frames available: 1440
...
0123:trace:xaudio2:IXAudio2Impl_CreateSourceVoice Created source voice:
0xfc8c9d8
0133:trace:xaudio2:do_engine_tick Calling OnVoiceProcessingPassStart with
BytesRequired: 19200
0123:trace:xaudio2:XA2SRC_SetVolume 0xfc8c9d8, 1.000000, 0x0
0133:trace:xaudio2:XA2SRC_GetState 0xfc8c9d8, 0x1643e908, 0x0
0133:trace:xaudio2:XA2SRC_GetState returning 0, queued: 0
0123:Call msvcr110.memcpy(0032eda0,0032ede0,0000000c) ret=0132da59
0123:Ret  msvcr110.memcpy() retval=0032eda0 ret=0132da59
0133:trace:seh:raise_exception code=c0000005 flags=0 addr=0x138e860 ip=0138e860
tid=0133
0133:trace:seh:raise_exception  info[0]=00000000
0133:trace:seh:raise_exception  info[1]=00000054
0133:trace:seh:raise_exception  eax=00000004 ebx=f7542000 ecx=00000054
edx=1643ef4c esi=14e30cb4 edi=1643fb40
0133:trace:seh:raise_exception  ebp=1643e918 esp=1643e8dc cs=0023 ds=002b
es=002b fs=0063 gs=006b flags=00210297 
...
--- snip ---

Slowing down by adding more trace or debugging avoids the crash sometimes but
most of the time it crashes immediately.

Reverting the patch or skipping it during build ('-W xaudio2_7-Abandon_Albufs')
avoids the crash and everything is fine.

Regards

-- 
Do not reply to this email, post in Bugzilla using the
above URL to reply.
You are receiving this mail because:
You are watching all bug changes.

[Bug 40963] Upwards, lonely robot Demo crashes with Too many buffers queued

https://bugs.winehq.org/show_bug.cgi?id=40963

Alistair Leslie-Hughes <leslie_alistair(a)hotmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
    Staged patchset|https://github.com/wine-com |
                   |pholio/wine-staging/tree/ma |
                   |ster/patches/xaudio2_7-Aban |
                   |don_Albufs                  |
             Status|STAGED                      |NEW

--- Comment #10 from Alistair Leslie-Hughes <leslie_alistair(a)hotmail.com> ---
Removing staging reference as it doesn't fix the issue at hand.  The original
patch by Andrew, even though has compile warnings corrects the issue.

The staging patch was changed thus still suffers "Too many buffers queued",
since it never updates nbufs which is the cause of the issue.

-- 
Do not reply to this email, post in Bugzilla using the
above URL to reply.
You are receiving this mail because:
You are watching all bug changes.

[Bug 40963] Upwards, lonely robot Demo crashes with Too many buffers queued

https://bugs.winehq.org/show_bug.cgi?id=40963

Andrew Eikum <aeikum(a)codeweavers.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
  Attachment #55163|0                           |1
        is obsolete|                            |

--- Comment #11 from Andrew Eikum <aeikum(a)codeweavers.com> ---
Created attachment 59811
  --> https://bugs.winehq.org/attachment.cgi?id=59811
xaudio2: Ignore buffers returned from OpenAL after Stop

Here's an improved version of the patch that addresses what Sebastian found.
Can you confirm it still fixes the issue, Alistair?

-- 
Do not reply to this email, post in Bugzilla using the
above URL to reply.
You are receiving this mail because:
You are watching all bug changes.

[Bug 40963] Upwards, lonely robot Demo crashes with Too many buffers queued

https://bugs.winehq.org/show_bug.cgi?id=40963

--- Comment #12 from Alistair Leslie-Hughes <leslie_alistair(a)hotmail.com> ---
(In reply to Andrew Eikum from comment #11)
> Created attachment 59811 [details]
> xaudio2: Ignore buffers returned from OpenAL after Stop
>
> Here's an improved version of the patch that addresses what Sebastian found.
> Can you confirm it still fixes the issue, Alistair?

Yes, the patch fixes the issue.

-- 
Do not reply to this email, post in Bugzilla using the
above URL to reply.
You are receiving this mail because:
You are watching all bug changes.

[Bug 40963] Upwards, lonely robot Demo crashes with Too many buffers queued

https://bugs.winehq.org/show_bug.cgi?id=40963

Alistair Leslie-Hughes <leslie_alistair(a)hotmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
      Fixed by SHA1|                            |e2c73fc394bace00c4463b6fcb6
                   |                            |dfe9dc64084bc
         Resolution|---                         |FIXED

--- Comment #13 from Alistair Leslie-Hughes <leslie_alistair(a)hotmail.com> ---
Fixed by 
http://source.winehq.org/git/wine.git/?a=commit;h=e2c73fc394bace00c4463b6fcb...

-- 
Do not reply to this email, post in Bugzilla using the
above URL to reply.
You are receiving this mail because:
You are watching all bug changes.

[Bug 40963] Upwards, lonely robot Demo crashes with Too many buffers queued

https://bugs.winehq.org/show_bug.cgi?id=40963

Alexandre Julliard <julliard(a)winehq.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |CLOSED

--- Comment #14 from Alexandre Julliard <julliard(a)winehq.org> ---
Closing bugs fixed in 3.0-rc1.

-- 
Do not reply to this email, post in Bugzilla using the
above URL to reply.
You are receiving this mail because:
You are watching all bug changes.

[Bug 40963] Upwards, lonely robot Demo crashes with Too many buffers queued

https://bugs.winehq.org/show_bug.cgi?id=40963

Michael Stefaniuc <mstefani(a)winehq.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |2.0.x

-- 
Do not reply to this email, post in Bugzilla using the
above URL to reply.
You are receiving this mail because:
You are watching all bug changes.

[Bug 40963] Upwards, lonely robot Demo crashes with Too many buffers queued

https://bugs.winehq.org/show_bug.cgi?id=40963

Michael Stefaniuc <mstefani(a)winehq.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|2.0.x                       |---

--- Comment #15 from Michael Stefaniuc <mstefani(a)winehq.org> ---
Removing the 2.0.x milestone from bugs included in 2.0.5.

-- 
Do not reply to this email, post in Bugzilla using the
above URL to reply.
You are receiving this mail because:
You are watching all bug changes.